Agent registry
A proposed inventory of supervised AI agents, their owners, and relevant operating context.
Source: ProductFeatureSetGlossary
Run timeline
A chronological view of an AI agent run and the events available for review.
Source: ProductFeatureSetTool-call log
A record of an attempted AI-agent tool action, its context, status, and related policy decision.
Source: ProductFeatureSetPolicy engine
The described component that evaluates an agent action against configured governance rules.
Source: ProductFeatureSetApproval queue
A review area where risky proposed actions wait for an authorized human decision.
Source: ProductFeatureSetRisk score
A contextual assessment used to support routing, blocking, sandboxing, or review.
Source: ProductFeatureSetQuality evaluation
A structured check of agent output or behavior against defined criteria.
Source: ProductFeatureSetIncident alert
A notice that an exception, policy issue, loop, or suspicious action needs attention.
Source: ProductFeatureSetAudit export
A portable record of available run, policy, approval, incident, and evaluation evidence.
Source: ProductFeatureSetPolicy version
A tracked edition of supervision rules so reviewers can identify what governed a run.
Source: ContextLayerTamper-evident log
A record designed to make later alteration detectable rather than silently rewriting history.
Source: TrustSafetyComplianceSafe fallback
A constrained response used when supervision data, permissions, or confidence are incomplete.
Source: AgenticAutonomyAndSafetyModelSandbox
A restricted operating state for examining an action without granting normal production reach.
Source: AgenticAutonomyAndSafetyModelBlocked action
A proposed agent action stopped under a configured policy or review boundary.
Source: ArtifactCaptureStrategyRun ingestion
The process of receiving agent-run events for timeline, policy, and audit review.
Source: MVP_ScopeSupervision control plane
The proposed shared layer for monitoring and governing multiple AI-agent workflows.
Source: SubcategoryIncident escalation
Routing a detected supervision problem to the accountable person or response process.
Source: TrustSafetyComplianceCost tracking
A proposed dashboard capability for observing usage-related measures across agent runs.
Source: ProductFeatureSet